[LLVMbugs] [Bug 21222] New: InstCombine: wrong folding of a constant comparison involving ashr and negative numbers.

[bugzilla-daemon at llvm.org]

http://llvm.org/bugs/show_bug.cgi?id=21222

            Bug ID: 21222
           Summary: InstCombine: wrong folding of a constant comparison
                    involving ashr and negative numbers.
           Product: libraries
           Version: trunk
          Hardware: PC
                OS: Windows NT
            Status: NEW
          Severity: normal
          Priority: P
         Component: Scalar Optimizations
          Assignee: unassignedbugs at nondot.org
          Reporter: andrea.dibiagio at gmail.com
                CC: llvmbugs at cs.uiuc.edu
    Classification: Unclassified

Reproducible:

;;
define i1 @test(i32 %B) {
  %shr = ashr i32 -93, %B
  %cmp = icmp eq i32 %shr, -2
  ret i1 %cmp
}
;;

Function @test should return true when %B is equal to 6.
That is because -93 >> 6 = -2.
Currently the Instruction Combiner wrongly folds the entire function to a
`return i1 false`.

The problem is in the folding logic located in InstCombineCompares.cpp; method
InstCombiner::FoldICmpCstShrCst wrongly computes the distance (in binary)
between two negative values.
With:
  AP1 < 0
  AP2 < 0
  AP2 != 0
  AP1 != 0
  AP1 > AP2

Currently the distance between AP2 and AP1 is wrongly computed as:
  (-AP2).logBase2() - (-AP1).logBase2()

the distance between AP1 and AP2 should be computed instead taking the ones'
complement of AP2 and AP1:
  (~AP2).logBase2() - (~AP1).logBase2()

This is another regression introduced by revision 213678 which, unfortunately
wasn't correctly fixed by revision 217950.

I have a fix for this bug which I plan to send soon for review on the mailing
list.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20141009/bda4d4e0/attachment.html>